The Relative Roles of Intraspecific Competition and Plant–Soil Feedback

Erschienen in: Oikos ; 125 (2016), 10. - S. 1458-1466 https://dx.doi.org/10.1111/oik.02770

Commonness and rarity of alien and native plant species – the relative roles of intraspeciﬁc competition and plant–soil feedback

Gregor Müller, Mark van Kleunen and Wayne Dawson

G. Müller (http://orcid.org/0000-0001-8859-0700)([email protected]), M. van Kleunen and W. Dawson, Ecology, Dept of Biology, Univ. of Konstanz, Universitätsstrasse 10, DE-78457 Konstanz, Germany. WD also at: School of Biological and Biomedical Sciences, Durham University, South Road, Durham, DH1 3LE, UK.

The success of invasive alien and common native species may be explained by the same underlying mechanisms. Differences in intraspecific competition as well as differences in plant–soil feedback have been put forward as potential determinants of plant success. We teased apart the relative roles of competition and plant–soil feedback in a greenhouse experiment with 30 common and rare alien and native species from nine plant families. We tested whether plant biomass decreased less for common than rare species, regardless of origin, when grown at higher relative frequencies (1, 3 or 6 out of 9 plants per pot) in a community and in soil previously conditioned by the same species at different frequencies (0, 1, 3 or 6 out of 9 plants per pot) in an orthogonal design for these two factors. Plant survival decreased slightly, but non-significantly, for all species when grown in soil previously occupied by conspecifics. Among surviving plants, we found a decrease in biomass with increasing intraspecific competition across all species (regardless of origin or commonness), and alien species were more negatively affected by previous high plant frequency than native species, but only marginally significantly so. Our findings suggest that, while intraspecific competition limits individual biomass in a density-dependent manner, these effects do not depend on species origin or commonness. Notably, alien species but not natives showed a decrease in performance when grown in soil pre-conditioned with a higher frequency of conspecifics. In conclusion, soil-borne pathogen accumulation might be weak in its effects on plant performance compared to intraspecific competition, with neither being clearly linked to species commonness.

Why are some species common while others are rare? This plant–soil feedback and intraspecific competition, and short question has puzzled ecologists for decades (Preston potential interactions between the two processes. Plant–soil 1948). With the emergence of invasion ecology in times of feedback has recently been proposed as a potential mecha- globalization (Mack et al. 2000, Bradley et al. 2010), the nism that could explain plant species commonness and rarity question ‘what determines species commonness’ has gained (Klironomos 2002, MacDougall et al. 2011, van der Putten further interest. Numerous theories and approaches that et al. 2013). Since plants influence their community of soil tackle this challenging question have been developed (Kunin biota and these in turn influence plant performance, such and Gaston 1993, Mitchell et al. 2006, Gaston 2011). How- host-specific plant–soil feedback may be an important regu- ever, general rules and the driving mechanisms behind some- lator of plant species abundance. In particular, differences times striking differences in species success have not been among species in accumulation of soil-borne pathogens, or clearly identified. The mechanisms explaining why some a low susceptibility to or even the absence of such pathogens alien species successfully spread and occupy large areas at could lead to the dominance or high abundance of common high abundances in the introduced range might be the same native and alien species. Especially invasive alien species as those explaining high abundance in their native range or might have left their soil pathogens behind, allowing them that similarly allow some native species to obtain a wide dis- to gain advantage over resident native species. An absence of tribution and a high abundance. This possibility has recently soil-borne pathogens would be in line with the enemy release received growing attention by ecologists (Thompson et al. hypothesis (ERH) (Keane and Crawley 2002), which so far 1995, van Kleunen et al. 2010). Thus, commonness and rar- has mainly been tested with regard to aboveground enemies ity of alien and native species might represent ‘two sides of (Mitchell and Power 2003, Liu and Stiling 2006). the same coin’ (Jeschke and Strayer 2008). Based on the theoretical framework of species coexistence, Differences in density-dependent enemy attack or self-limitation of species by intraspecific competition should resource partitioning have been put forward to explain be stronger than limitation by interspecific competition plant species success (Adler et al. 2007, MacDougall et al. providing a stabilizing mechanism that allows for species 2009). We specifically focus on two processes in our study; coexistence as lined out by (Chesson 2000). However, species

1458 Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-0-327214 might differ in the magnitude of the difference between self- and thus increase the negative impact on per capita plant limitation and limitation by other competitors (Comita performance? et al. 2010). In other words, common species might be less self-limited in terms of intraspecific competition than rare species. Material and methods If plant–soil feedback acts in a density-dependent manner, with more negative effects of soil biota when conspe- In order to be able to generalize results beyond a few study cifics grow at high density, then plant–soil feedback and species (van Kleunen et al. 2014) we conducted a multi-spe- self-limitation might operate simultaneously and may inter- cies greenhouse experiment using 30 different target species act to determine plant performance. The relative importance from nine different families (Table 1). The chosen target spe- of the two mechanisms in explaining success of alien and cies represented taxonomic groups, which ideally contained native species has yet to be tested. Consequently, testing both at least one common native, one rare native, one common factors simultaneously may shed light on whether an interac- alien and one rare alien plant species. As a proxy for the tion between plant–soil feedback and intraspecific competi- degree of commonness of each species, we used the number tion takes place. For example Petermann et al. (2008) stated of 130-km2 grid cells occupied by the species in Germany that negative plant–soil feedback might hamper the compet- (maximum 3000) extracted from the FloraWeb database itive ability of plants (see also Kardol et al. 2007) leading to (FloraWeb, Bundesamt für Naturschutz). We also aimed the possibility that plant–soil-feedback effects become more to choose species that occur in similar habitats, excluding important than self-limitation by intraspecific competition habitat specialists (excluding also woodland and riverine spe- in regulating plant species abundances. cies), and which are not simply rare due to their geographic To test the roles of plant–soil feedback and intraspe- distribution overlapping only marginally with the borders cific competition in explaining species success, we used a of Germany. Alien species were defined as being introduced greenhouse experiment with 30 herbaceous plant species, to Germany after 1492. Another three species from fami- including 13 native and 17 alien species, respectively, that lies different to the ones of the target species, Phleum prat- vary in commonness within Germany. We grew the spe- ense (2558 grid cells), Galium album (2463 grid cells), and cies at different intraspecific frequencies and in soils previ- Ranunculus acris (2985 grid cells), were selected to represent ously occupied by the same species at differing frequencies, a common background community typical for grasslands in and analysed individual plant performance. We asked the Germany. The experiment consisted of two growing phases. following questions: 1) Are common species less nega- Phase one was used as a soil-conditioning phase to build up a tively affected by density-dependent plant–soil feedback potentially species-specific microbial community. Phase two than rare species, irrespective of origin? 2) Are common was then used for testing the effect of increasing intraspecific species less self-limited by intraspecific competition than competition under simultaneous influence of the pre-trained rare species, irrespective of origin? 3) Do plant–soil feed- soil microbial community at different frequencies, allow- backs and intraspecific competition interact synergistically ing for an orthogonal study design with respect to phase 1

Table 1. The 30 study species and their respective commonness (measured as number of ^130 km2 grid cells occupied in Germany (out of 3000 maximum), (FloraWeb, Bundesamt für Naturschutz)) listed by family and origin. Year of introduction of alien species in Germany as found in Krausch (2003) and year of naturalization (FloraWeb, Bundesamt für Naturschutz) are displayed. The percentage of pots per species (out of 36 pots maximum) that was not affected by mortality in phase 2 is also shown. Species in bold font were represented in all treatment combinations.

Alien Native Grid Year introduced % of Grid % of Family Species cells (naturalized) pots Species cells pots Apiaceae Anethum graveolens 576 () 94.4% Daucus carota 2942 88.8% Foeniculum vulgare 297 () 86.1% Oenanthe lachenalii 104 75% Asteraceae Artemisia verlotiorum 168 (1920) 33.3% Achillea millefolium 2741 80.5% Aster novi-belgii 1530 1686 (18th cent.) 88.9% Gnaphalium luteoalbum 562 44.4% Solidago canadenis 2660 1648 (1736) 83.3% Lactuca saligna 119 88.8% Senecio jacobaea 2773 61.1% Brassicaceae Cardaria draba 1513 (1728) 36.1% Cardamine pratensis 2923 41.6% Lepidium heterophyllum 98 () 86.1% Lepidium graminifolium 86 88.8% Caryophyllaceae Cerastium tomentosum 1296 1594 (16th cent.) 80.5% Silene latifolia 2893 80.5% Gypsophila paniculata 122 1757 () 86.1% Geraniaceae Geranium macrorrhizum 146 1588 () 27.7% Geranium lucidum 103 80.5% Geranium pyrenaicum 2134 17th cent. (1800) 91.6% Geranium pratense 1844 66.6% Lamiaceae Salvia pratensis 1694 41.6% Plantaginaceae Linaria dalmatica 21 () 63.8% Pseudolysimachion spicatum 559 63.8% Veronica persica 2863 (1805) 91.6% Veronica chamaedrys 2913 80.5% Digitalis lutea 127 47.2% Scrophulariaceae Scrophularia vernalis 105 18th cent. (1768) 58.3% Phrymaceae Mimulus guttatus 694 1812 (1830) 36.1%

1459 frequencies (i.e. plant-soil feedback) and phase 2 frequencies agar plates that were inoculated with the substrate showed (i.e. intraspecific competition. no visible signs of contamination. We planted the species in 4.2-1 square pots (I6 X I6 X I6.4 em) filled with -4 I Experimental setup phase 1 mixed substrate. The substrate consisted of 250 ml field soil inoculum, mixed with I: I vermiculite:sterile sand and 20 g We planted a total of nine plants per pot in a 3 X 3 square 8- 9 month slow release fertiliser pellets (NPK = I8:6:I2). (Fig. I). For the target species we planted zero, one, three Because of the large size of the experiment, we planted the or six plants per pot. The remaining plants planted were the species over a staggered period between 27 August and 7 three community species. These were planted using the fol September 20 I2. We replaced dead plants within a period of lowing frequencies: pots with zero target plants contained two weeks after initial planting. The pots were blocked into six Ph/cum pratense, two Galium album and one Ranunculus the three greenhouses (one replicate in each) and were ran acris plants; pots with one target plant contained five Phleum domized within the greenhouses. We set the mean tempera pratense, two Galium album and one Ranunculus acris plants; ture to 2I °C during the day and I6°C at night. Lights were pots with three target plants contained four Phleum pratense, switched on for I4 h per day but only if natural light levels one Galium album and one Ranunculus acris plants; pots with fell below IOO Jlmol s-1 m-2. After six weeks, we moved a six target plants contained two Ph/cum pratense, and either randomly chosen subset of the pots into a fourth greenhouse one Galium album or one Ranunculus acris plants chosen at compartment with the same growing conditions in order to random. We chose these frequencies in order to maintain provide sufficient space for continued growth. We watered a ratio of approximately I :2 of forbs to grasses among the the plants once every three days to levels that ensured no lack community species. We replicated each frequency level nine of water availability. times in the first phase resulting in 36 pots for each target We harvested the aboveground biomass of plants of the species including the nine 'community' pots, which did not first phase after a growing period of eight weeks. We then contain a target plant. The total number of pots in the first dried (72 hat 80°C), and weighed the biomass and calcu phase consequently was I 080. lated the per capita aboveground biomass (total target bio Prior to the start of the first growing phase, we collected mass divided by number of target plants in the pot).After the approximately 250 I of soil from six grassland areas in the harvest, we sieved and homogenised the substrate in the pots vicinity of the Univ. ofKonstanz (list of species occurring at through a 5 mm mesh, removed roots and rhizomes from the the site is given in Supplementary material Appendix I Table soil, and put the substrate back in its original pot. All pots AI). We systematically took 5- I5 samples (IO em deep) per were then returned to their original greenhouse compart area along transects with a spacing of approximately I 0 m ments and stored at an air temperature of 5- 8°C until they between each sampling point. We pooled the samples and were required for the second phase of the experiment. homogenised the soil by sieving through a 5 mm mesh to remove roots, stones and other plant material. This soil was Experimental set-up phase 2 then used as a soil inoculum for each pot. The species were germinated in a growth chamber We germinated the plants for the second phase in January (temperature = I5°C/20°C, I2 h!I2 h darkness/light, light 20I3 under similar growth conditions as the seedlings for level = I 50 Jlmol m- 2s- 1, relative humidity = 90%) on a I :I phase I, and planted them again staggered from 28 January sand:vermiculite substrate. The substrate was not sterilized to I 0 February 20 I3. We planted the target species always prior to the germination of seedlings, however, incubated in pots that previously contained the same species or only Phase 1 mmmm 1 1 1 1 Phase 2 mmmm

0 Community species • Target species mmmm mmmm Figure 1. Schematic illustration of the experimental setup. Phase 1 represents the soil conditioning phase. Phase 2 represents varying intraspecific competition levels in preconditioned soil of phase 1. (Replicated three times for each species).

I460 community plants in the following manner: target species in planting frequency in phase 2 using the ‘multcomp’ package frequencies of one, three and six out of nine were planted in (Hothorn et al. 2008). pots previously containing zero, one, three, or six conspecific For the biomass analysis, we only used the subset of pots plants (i.e. in phase 1). This resulted in three pots per target in which all target plants survived, which resulted in a data species for each combination of planting frequency in the set of 719 pots (out of 1056 pots). All 30 species were rep- second phase and planting frequency of the first phase (Fig. resented in this subset. We used linear mixed effect models 1). Thus, we achieved a fully orthogonal design. We filled in the lme4 package to analyse per capita aboveground bio- the remaining positions in the pots again with community mass. Per capita aboveground biomass (in grams) was natural species in the same way as in phase 1. Because of variation log-transformed prior to analysis to achieve normality of the in substrate volumes due to loss from pots during sieving, residuals. The fixed and random effects were the same as in we placed 1 l of 1:1 vermiculite and sterile sand mixture at the model used for the analysis of survival. Similarly we used the bottom of each pot and refilled the remainder with the stepwise backward model selection via likelihood-ratio tests substrate of phase 1. to assess significance of the model terms. The ‘multcomp’ Since Oenanthe lachenalii germinated in insufficient package was used to test for differences among levels of phase numbers, we only planted this species at phase 2 frequencies 1 and phase 2 planting frequencies. Furthermore, to ensure of one and six out of nine plants in pots with soil of one and that our results were not affected by species that were absent six plants in phase 1. Thus we finally had a set of 29 species in some treatment combinations, we analysed a subset of with 36 pots each, representing three replicates of all respec- the data excluding those species (Table 1). The analysis was tive combinations of phase 1 and phase 2, and one species performed in the same way as for the complete data set. To with twelve pots, making a total of 1056 pots in phase 2. assess whether relationships between per capita biomass and We kept the plants under the same greenhouse conditions commonness were non-linear, we also performed the same as in phase 1, and applied the same watering regime. To analysis with a discretized commonness variable (rare; inter- reduce mortality after planting, we delayed additional light- mediate; common – based on clear groupings evident in the ing until 25 February. We then increased lighting from five Supplementary material Appendix 1 Fig. A4). However, per to eight and finally to 14 h in a two-day stepwise interval. capita performance was not significantly explained by this We replaced dead plants within a period of two weeks from discrete measure of commonness and hence, the minimum initial planting. model remained the same (data not shown). After a growth period of 10 weeks, we harvested the plants, again in a staggered manner following their planting sequence (8–19 April 2013). We dried and weighed the Results aboveground biomass of all plants following the same proce- dure as in phase 1. We counted and recorded the number of Survival response surviving plants one week before the harvest. Analysis of survival showed that none of the model terms was Analyses significant (Supplementary material Appendix 1 Table A2). Increasing phase 1 planting frequency resulted in slightly Due to high mortality of target plants in phase 2, we split lower survival for all plant species, however, this effect was the statistical analysis into an analysis of survival, and, for only marginally significant and accounted only for a 3% the subset of plants that survived, an analysis of aboveground lower survival probability between the different planting biomass. We used the proportional data on survival of target frequencies (Fig. 2). plants per pot to analyse probability of survival per target. Survival was analysed using a generalised linear mixed model Biomass responses with binomial error distribution in the lme4 package (Bates et al. 2014) in the software R ver. 3.1.1. We used the opti- The minimum model for per capita biomass retained phase mizer ‘bobyqa’ and set the maximum number of iterations 2 planting frequency as a significant main effect and a to 100 000 to achieve model convergence. Species nested significant interaction between species origin and phase in family and greenhouse compartment were included as 1 planting frequency (Table 2, Supplementary material random effects. Initial phase 1 planting frequency (0, 1, 3, Appendix 1 Table A3). Per capita biomass of the target 6 out of 9 plants per pot, i.e. plant–soil feedback effect), species was reduced by increasing intraspecific plant fre- initial phase 2 planting frequency (1, 3, 6 out of 9 plants quency in phase 2 (Fig. 3). Multiple comparisons between per pot, i.e. intraspecific competition), commonness as a phase 2 planting frequencies revealed that pots with six continuous variable (number of grid cells occupied by the target plants in phase 2 showed a significant reduction in species in Germany; centred on the mean and scaled by per capita biomass compared to pots with one target plant the standard deviation), origin and all respective interac- in phase 2 (mean difference –0.209, 95% CI –0.376; tions were included as fixed effects in a four-way interaction –0.041, p 0.004, Fig. 3). Reductions in per capita biomass model. We also added total biomass per pot in phase 1 as a of pots with six target plants compared to pots with three covariate (centred to the mean and scaled by the standard target plants in phase 2 (mean difference –0.150, 95% deviation). We used stepwise backward model selection via CI –0.324; 0.017, p 0.146, Fig. 3) and of pots with three likelihood-ratio tests to obtain a minimum model and to targets compared to pots with one target in phase 2 were not test for significance of interactions. We performed multiple significant (mean difference –0.058, 95% CI –0.214; pairwise comparisons to test for differences among levels of 0.097, p 0.956, Fig. 3).

1461

(a) (b)

6.0 6.0

3.0 3.0

1.5 1.5

0.8 0.8

0.4 0.4 Per capita biomass in [g] Per

0.2 0.2

0 0 0 1 3 6 0 1 3 6 Frequency of target plants in phase 1

Figure 4. Panel (a) displays least-square mean per capita biomass ( SE) of 17 alien species grown in soil previously occupied by 0, 1, 3 or 6 target plants of the same species in phase 1 (averaged across target plant frequencies in phase 2). Black dots display means across the 17 target species. Grey dots indicate the respective means for each of the 17 target species. Panel (b) displays least-square mean per capita biomass ( SE) of 13 native species grown in soil previously occupied by 0, 1, 3 or 6 target plants of the same species in phase 1 (averaged across target plant frequencies in phase 2). Filled white dots display means across the 13 target species. Grey dots indicate the respective means for each of the 13 target species. (y-axis is displayed on natural log scale in both panels). not driven by underrepresented species in our data set. To a prevalence of negative plant–soil feedbacks for the focal the contrary, this analysis revealed a more significant effect species in semiarid grasslands, but did not find a clear pat- of phase 1 planting frequencies for alien species, and non- tern explaining species abundances with regard to effects of significant effects for native species (Supplementary material soil biota. Our findings similarly point towards a limited Appendix 1 Fig. A7). role of plant–soil feedback in explaining species commonness, and plant–soil feedback effects may rather be largely context dependent (Bezemer et al. 2006, Reinhart 2012). Discussion An explanation for our findings, specifically the absence of a significant and concordant pattern in reduction of In this study, we tested if differences in intraspecific compe- plant performance by plant–soil interactions, likely lies in tition and plant–soil feedback could explain commonness the experimental approach that we used. We did not test of alien and native species. We found that per capita plant plant performance in pre-trained soils in comparison to performance declined with increasing intraspecific plant fre- sterilized soils, which is an often-used method in plant–soil quency, regardless of origin and commonness, which sug- feedback experiments and which leads to stronger feedback gests that alien and native, and rare and common species effects (van der Putten and Peters 1997, van der Putten are affected similarly by intraspecific competition. Negative et al. 2007, MacDougall et al. 2011, but see Brinkman plant–soil feedback effects were relatively weak, and alien et al. 2010). Instead we used a set of common grassland but not native species showed a reduction of performance species as a neutral community, creating a much more real- when grown in soils previously occupied by the same species istic soil-microbiome control. Consequently, the ‘neutral’ at high frequencies. Furthermore, we did not find evidence community could have accumulated already a high density for interactions between density-dependent soil effects and of pathogens so that native species experienced a ‘ceiling- intraspecific competition. effect’, such that an increase in frequency of a target species In contrast to expectations, our study shows that grow- does not lead to a substantial further reduction of biomass ing plants in soil previously occupied by the same species due to the presence of pathogens. Similarly, Maron et al. at increasing frequencies did not lead to a stronger reduc- (2014) found that negative soil effects for species grown tion of performance of rare species compared to common in soil with their native soil biota seem to develop largely species. However, alien plant species showed reduced per independently of study species presence. This would indi- capita biomass when grown in soil conditioned by the cate that generalists among soil pathogens might play an highest conspecific plant frequency compared to control important role. We also found a marginally non-significant community soil, whereas native species were generally not trend that phase 1 planting frequencies cause mortality in affected by phase-1 planting frequencies. In a previous phase 2, albeit with a very small effect size that cannot fully study Klironomos (2002) found evidence for consistently explain greater mortality rate observed in the second phase positive plant–soil feedback for common alien species com- of the experiment compared to the first. Mortality also did pared to a consistently negative feedback of rare native spe- not differ according to species origin or commonness. This, cies. In a comparable experiment, Reinhart (2012) found combined with the overall greater mortality in the second

1463 phase may further indicate the accumulation of generalist performance more negatively than interspecific competi- pathogens within the whole experiment. tion (Chesson 2000). Despite its clear theoretical under- The limited role of plant–soil feedback that we find in our pinning, empirical proof of this concept has been equivocal study does not mean that for a particular species in a partic- (Goldberg and Barton 1992, Gurevitch et al. 1992, but see ular context, plant–soil-feedback effects are not important. Levin and HilleRisLambers 2009). Our study supports the There are a number of studies that find negative feedback idea of stronger self-limitation, but we did not find evidence effects on plant performance (Kulmatiski et al. 2008) that for any differences in density-dependent intraspecific com- are likely driven by species-specific pathogens (Bezemer et al. petition between alien and native species, or in relation to 2006, Kardol et al. 2007, Petermann et al. 2008, Reinhart commonness. However, it is important to note that inter- 2012). However, the complexity of plant–soil interactions specific effects could have an influence on the performance and their dependence on multiple factors (i.e. soil type, soil- of the focal species in species mixtures, as other studies show legacy history, presence of competitors) often results in large (Bezemer et al. 2006, Kardol et al. 2007). Kardol et al. variation among species and study systems (Kulmatiski et al. (2007) for example reported that a selection of early suc- 2008), thus preventing clear, general patterns from being cessional plant species showed differential responses towards observed among species. heterospecific soil inocula, ranging from positive to negative. An alternative explanation for our results may be that pot In our case this specifically means that interspecific effects of limitation (Poorter et al. 2012) has magnified the effect of the community species, e.g Phleum pratense, may influence competition relative to plant–soil-feedback effects, so that our results besides intraspecific effects of the targets. Due effects of the soil biota were overridden by competition for to the design of our study, which aimed to entail a realis- root space. However, growing plants in larger pots would tic grassland background community, we cannot fully assess have reduced the potential for plants to compete, and plants this role of interspecific effects. Nonetheless, Blank (2010) may also experience intense belowground competition in also reports on stronger effects of intraspecific competition natural communities (Casper and Jackson 1997). Thus, compared to interspecific competition for a set of native and we consider the conditions under which competition and alien species, but highlights that alien species might gain plant–soil feedback can be detected in our experiment to be advantage over natives by better capitalizing on nutrients in reasonably realistic. highly fertile soils (see also Dawson et al. 2012). This may Nevertheless, alien species in our study showed a reduc- also explain the higher per capita performance of aliens com- tion in performance when grown in soil previously occupied pared to natives in our study, since plants were grown with by conspecifics at high frequencies. These findings may be addition of slow-release fertilizer and should therefore not explained by the nature of the interactions between alien have been limited in nutrient supply. Furthermore, Duralia species and their new soil biota. Alien species might on the and Reader (1993) tested if abundance of three prairie grasses one hand not be affected by some of the soil-borne pathogens is explained by competitive ability in a replacement series of their new range and might even have left some of their co- experiment, and found only weak evidence for a relationship evolved enemies of their native range behind. On the other between commonness and competitive ability. Despite the hand they might be naïve towards some of their soil-borne theoretical importance of density-dependent self-limitation enemies in the new range resulting in accordingly strong det- in regulating species abundance and coexistence, we found rimental effects (Parker et al. 2006, Parker and Gilbert 2007, no evidence that it covaries with commonness of either alien Verhoeven et al. 2009). Verhoeven et al. (2009) argue that or native species. ‘novelty’ can be claimed for both sides of the interaction, the Although plant–soil feedback has been proposed as a plant as well as the pathogen. Consequently there can be a mechanism that could drive species success (Klironomos mismatch that leads to enemy release, but also a mismatch 2002, van der Putten et al. 2013), we found no evidence that that leads to biotic resistance, which may explain the unex- commonness is explained by differences in density-depen- pected divergent plant–soil feedback effects on alien and dent plant–soil feedback. A meta-analysis on the effects native species in our study. of plant–soil feedback by Kulmatiski et al. (2008) showed However, since the selected species in our experiment that there is a general signal for a reduction in plant per- have been present in Germany for at least two centuries, alien formance due to plant–soil feedback. However, Kulmatiski species and their respective pathogens might have already et al. (2008) as well as van de Voorde et al. (2012) and Brink- adapted, resulting in the reduced performance of the alien mann et al. (2010) raise the point that varying experimental species that we observe in our experiment. For example, Diez protocols and a bias towards simplified greenhouse studies, et al. (2010) found that the negative plant–soil-feedback each with a limited but different set of target species and life effect of alien species in New Zealand increased with increas- forms, may account for a considerable amount of variation ing residence time. In contrast, Speek et al. (2015) did not in study outcomes. Another aspect in studies on plant–soil find such a pattern among alien species in a multi-species feedback is that only net outcomes of plant–soil interactions study in the Netherlands. These contrasting findings reflect are measured (e.g. biomass), however, this overall perfor- the complexity of plant–pathogen interactions, such that mance results from potentially multiple antagonistic (e.g. changes in the effects of interactions over time are unlikely pathogens) and mutualistic (e.g. mycorrhiza) interactions to be consistent. and physical properties of the soil (Reinhart and Callaway We found that an increase in current conspecific fre- 2006, van der Putten et al. 2013). Disentangling the relative quency led to a strong reduction in per capita biomass. This contributions of antagonistic and mutualistic soil organisms finding is in line with the predictions of coexistence theory, under controlled conditions will provide important insights namely that intraspecific competition should affect species into the underlying mechanisms. Moreover, transferring

1464 these insights to manipulative experiments under realistic Adler, P. B. et al. 2007. A niche for neutrality. – Ecol. Lett. 10: field conditions with a focus on population dynamics cover- 95–104. ing the whole life cycle of a study organism (Maron et al. Bates, D. et al. 2014. lme4: linear mixed-effects model using Eigen 2010, Flory and Clay 2013) may lead the way to a better and S4. R package ver. 1.1-7. – http://CRAN.R-project.org/ package lme4 . understanding of plant–soil interactions. Blank, R. R. 2010. Intraspecific and interspecific pair-wise seedling A specific aspect that arises by studying the role of plant– competition between exotic annual grasses and native peren- soil interactions in driving species commonness is the two- nials: plant–soil relationships. – Plant Soil 326: 331–343. way nature of species commonness in this relationship. Bezemer, M. T. et al. 2006. Plant species and functional group On the one hand, commonness may be the outcome or in effects on abiotic and microbial soil properties and plant–soil part be influenced by plant–soil interactions, while on the feedback responses in two grasslands. – J. Ecol. 94: other hand the commonness of a species is also a factor that 893–904. determines the chances for pathogens and other soil biota Bradley, B. A. et al. 2010. Predicting plant invasions in an era of global change. – Trends Ecol. Evol. 25: 310–318. to encounter (and potentially adapt to) their host or vice Brinkman, E. P. et al. 2010. Plant–soil feedback: experimental versa. Hence, commonness is a dependent and indepen- approaches, statistical analyses and ecological interpretations. dent variable at the same time. Furthermore the chances of – J. Ecol. 98: 1063–1073. encountering a pathogen (or a host) also depend on disper- Bundesamt für Naturschutz, FloraWeb. – www.floraweb.de sal abilities of both parts of the pathogen–plant interaction, Accessed 16 February 2014. which adds spatial and temporal dimensions to plant–soil Casper, B. B. and Jackson, R. B. 1997. Plant competition under- interactions (Adler and Muller-Landau 2005). Considering ground. – Annu. Rev. Ecol. Syst. 28: 545–570. these spatial and temporal dimensions may help to improve Chesson, P. 2000. Mechanisms of maintenance of species diversity. – Annu. Rev. Ecol. Syst. 31: 343–366. our understanding of whether plant–soil feedbacks drive the Comita, L. S. et al. 2010. Asymmetric density dependence shapes patterns of commonness we observe in both alien and native species abundances in a tropical tree community. – Science plant species. 329: 330–332. Dawson, W. et al. 2012. Common and rare plant species respond Conclusions differently to fertilisation and competition, whether they are alien or native. – Ecol. Lett. 15: 873–880. Our study suggests that plant–soil feedback plays little role Diez, J. M. et al. 2010. Negative soil feedbacks accumulate over in determining the success of common native and alien spe- time for non-native plant species. – Ecol. Lett. 13: 803–809. cies compared to rare natives and aliens, at least for a number Duralia, T. E. and Reader, R. J. 1993. Does abundance reflect competitive ability – a field-test with 3 prairie grasses. – Oikos of species from several families in central Europe. Further- 68: 82–90. more, species responses to increasing intraspecific frequency Flory, S. L. and Clay, K. 2013. Pathogen accumulation and long- are not linked to commonness and origin, suggesting that term dynamics of plant invasions. – J. Ecol. 101: 607–613. intraspecific competitive ability does not in general drive Gaston, K. J. 2011. Common ecology. – Bioscience 61: 354–362. alien and native plant species commonness or rarity. Con- Goldberg, D. E. and Barton, A. M. 1992. Patterns and conse- trary to the notion of enemy release, we also found that alien quences of interspecific competition in natural communities species suffer more from plant–soil feedback than native spe- – a review of field experiments with plants. – Am. Nat. 139: cies. Rather than being released from soil-borne pathogens, 771–801. Gurevitch, J. et al. 1992. A meta-analysis of competition in field alien species may suffer more from novel belowground inter- experiments. – Am. Nat. 140: 539–572. actions; alternatively, adaptation of pathogens to the alien Hothorn, T. et al. 2008. Simultaneous inference in general para- species may have already taken place, resulting in negative metric models. – Biometrical J. 50: 346–363. feedback. Thus, enemy release needs to be interpreted as a Jeschke, J. M. and Strayer, D. L. 2008. Are threat status and potentially dynamic process since a static interpretation may invasion success two sides of the same coin? – Ecography 31: not capture the evolutionary aspects in the plant–pathogen 124–130. interaction. While we found no clear role for plant–soil feed- Kardol, P. et al. 2007. Microbe-mediated plant–soil feedback causes back in explaining current commonness of alien and native historical contingency effects in plant community assembly. – Ecol. Monogr. 77: 147–162. species, future studies should consider whether plant–soil Keane, R. M. and Crawley, M. J. 2002. Exotic plant invasions feedback is linked to dynamic changes in species common- and the enemy release hypothesis. – Trends Ecol. Evol. 17: ness over time. 164–170. Klironomos, J. N. 2002. Feedback with soil biota contributes to Acknowledgements – GM and WD were funded by the Deutsche plant rarity and invasiveness in communities. – Nature 417: Forschungs Gemeinschaft (Project DA 1502/1-1). We thank O. 67–70. Ficht, Tilman Rönneburg, Sina Glöckner and Anna Gorenflo for Krausch, H.-D. 2003. Kaiserkron und Päonien rot… Entdeckung horticultural assistance and practical help. Thanks also go to the und Einführung unserer Gartenblumen. – Dölling und Galitz International Max Planck Research School for Organismal Biology Verlag. for supporting GM. Kulmatiski, A. et al. 2008. Plant–soil feedbacks: a meta-analytical review. – Ecol. Lett. 11: 980–992. Kunin, W. E. and Gaston, K. J. 1993. The biology of rarity – References patterns, causes and consequences. – Trends Ecol. Evol. 8: 298–301. Adler, F. R. and Muller-Landau, H. C. 2005. When do localized Levine, J. M. and HilleRisLambers, J. 2009. The importance of natural enemies increase species richness. – Ecol. Lett. 8: niches for the maintenance of species diversity. – Nature 461: 438–447. 254–257.

1465 Liu, H. and Stiling, P. 2006. Testing the enemy release hypothesis: Preston, F. W. 1948. The commonness, and rarity, of species. a review and meta-analysis. – Biol. Invas. 8: 1535–1545. – Ecology 29: 254–283. MacDougall, A. S. et al. 2009. Plant invasions and the niche. – J. Reinhart, K. O. 2012. The organization of plant communities: Ecol. 97: 609–615. negative plant–soil feedbacks and semiarid grasslands. MacDougall, A. S. et al. 2011. Weak conspecific feedbacks and – Ecology 93: 2377–2385. exotic dominance in a species-rich savannah. – Proc. R. Soc. Reinhart, K. O. and Callaway, R. M. 2006. Soil biota and invasive B 278: 2939–2945. plants. – New Phytol. 170: 445–457. Mack, R. N. et al. 2000. Biotic invasions: causes, epidemiology, Speek, T. A. A. et al. 2015. Local dominance of exotic plants global consequences and control. – Ecol. Appl. 10: 689–710. declines with residence time: a role for plant–soil feedback? Maron, J. L. et al. 2010. Using experiments, demography and – AoB Plants 7: plv021. population models to estimate interaction strength based on Thompson, K. et al. 1995. Native and alien invasive plants: more transient and asymptotic dynamics. – J. Ecol. 98: 290–301. of the same? – Ecography 18: 390–402. Maron, J. L. et al. 2014. Invasive plants escape from suppressive van de Voorde, T. F. J. et al. 2012. Soil inoculation method deter- soil biota at regional scales. – J. Ecol. 102: 19–27. mines the strength of plant–soil interactions. – Soil Biol. Mitchell, C. E. and Power, A. G. 2003. Release of invasive plants Biochem. 55: 1–6. from fungal and viral pathogens. – Nature 421: 625–627. van der Putten, W. H. 1997. Plant–soil feedback as a selective force. Mitchell, C. E. et al. 2006. Biotic interactions and plant invasions. – Trends Ecol. Evol. 12: 169–170. – Ecol. Lett. 9: 726–740. van der Putten, W. H. and Peters, B. A. M. 1997. How soil-borne Parker, I. M. and Gilbert, G. S. 2007. When there is no escape: pathogens may affect plant competition. – Ecology 78: the effects of natural enemies on native, invasive and nonin- 1785–1795. vasive plants. – Ecology 88: 1210–1224. van der Putten, W. H. et al. 2013. Plant–soil feedbacks: the past, Parker, J. D. et al. 2006. Opposing effects of native and exotic the present and future challenges. – J. Ecol. 101: 265–276. herbivores on plant invasions. – Science 311: 1459–1461. van Kleunen, M. et al. 2010. Are invaders different? A conceptual Petermann, J. S. et al. 2008. Janzen–Connell effects are widespread framework of comparative approaches for assessing determi- and strong enough to maintain diversity in grasslands. – Ecol- nants of invasiveness. – Ecol. Lett. 13: 947–958. ogy 89: 2399–2406. van Kleunen, M. et al. 2014. The more the merrier: multi-species Poorter, H. et al. 2012. Pot size matters: a meta-analysis of the experiments in ecology. – Basic Appl. Ecol. 15: 1–9. effects of rooting volume on plant growth. – Funct. Plant Biol. Verhoeven, K. J. F. et al. 2009. Plant invaders and their novel 39: 839–850. natural enemies: who is naive? – Ecol. Lett. 12: 107–117.

1466